Using 3D CAD Models For Value Visualization: An Approach With SIEMENS .

4Fig. 2: The value assessment process [8]In order to configure the value models, the design team needs to retrieve information about therelationship between drivers and objectives, for instance how MTBF affects Availability in real life(Phase 2). Furthermore, because some aspects of the product lifecycle may be more important thenothers, the team must assign weights to each driver. It is also important to notice that the informationthat characterizes such value models features different levels of confidence (some relationships mightbe supported by experimental evidence, others might be just educated guesses). In parallel, designersdevelop handful solution concepts, using as input the available list of requirements and value drivers,which are formalized within the CAD/PLM environment.During Phase 3 the value models are fed with value drivers and objectives, and computed toproduce a ranking of the current designs, highlighting negatively impacted areas and suggestingnecessary corrective actions (for a more detailed description of the value model adopted in the casestudy see: [7]). These results are communicated to the designers, who update the designs consideringboth the requirement description and the value information (Phase 4). If a value dimension is belowthe acceptance criteria, it is discussed within the team to implement the necessary corrective actions,such as modifying the product geometry, introducing a new material or involving external resources tosupport the development work. The designs are then re-assessed and updated. The process is iterateduntil a satisfactory value is found. The number of iterations depends on the complexity of the productand on time constraints.In the Integrated Analysis step (Phase 5), the team compiles all the material needed at the gate.The final value models are computed and included in a Value Report, which also includes feedbackabout the level of maturity/fidelity of the models used for the value computation.At the gate meeting (Phase 6), the decision material is reviewed, a questions and answers sessionwith the project leader is performed, and a decision is made about the continuation of the project. Thediscussion aims at resolving trade-offs between alternative concepts, focusing on areas where valuecontribution is perceived as weak. This session focuses both on the numbers (i.e., value) and onmaturity of the knowledge behind the numbers. Where needed, additional value analyses arerequested to verify the correctness of the value statement and to decide among the trade-offs.Eventually, the gate is opened, the expectations for the next gate are communicated to the projectleader, the acceptance criteria for the next gate are decided and resources allocated.The development of the color-coded 3D CAD models approach for value visualization targetsPhase 4 and 6 of the process pictured in Fig.2. The decisions taken in these phases, in fact, result fromthe team members’ debate around which value drivers to prioritize, which lifecycle aspects to improve,which engineering characteristics to prefer. The ability to communicate the value of aproduct/technology along multiple criteria is crucial to stimulate such early stage discussions and togrow a common understanding of “value provision”, eventually helping the team in making a moresound assessment on the design that is best aligned with the purpose of the project.Computer-Aided Design & Applications, 10(a), 2013, bbb-ccc 2013 CAD Solutions, LLC, http://www.cadanda.com

52.2The Value Assessment ModelIn Phase 3, the value associated to the IMC concept was calculated using an ad-hoc value assessmentmodel named Early Value-Oriented design exploration with KnowlEdge maturity (EVOKE) [7]. EVOKE isderived from the Customer Oriented Design Analysis (CODA) [40] matrix, which, in turn, is anextension of Quality Function Deployment (QFD) [2]. EVOKE takes as input the existing objectives(engineering characteristics) for a design concept (e.g., shape, size, material) together with the list ofvalue drivers (e.g., manufacturability or availability) to produce a score (Design Merit or DM) expressingthe “goodness” of the design through a percentage score from 1% to 100%. EVOKE uses non-linearoptimization type functions to map objectives to value drivers (Fig. 3). This is believed to betterapproximate the customer response to changes in a product attribute [3].Fig. 3: Non -linear optimization type functions used in EVOKETable 1 shows an extract of the EVOKE matrix used to assess the IMC options in the study.Initially, the design team models the relationship between an objective and a driver using correlationcoefficients. In the example, Surface finishing (the objective) is linked to Drag, Manufacturability andKnowledge Reuse (the value drivers) by strong (0.9), medium or weak correlations (0.1). A RelationshipType further characterizes this link: Drag is improved when the friction coefficient is minimized, whileManufacturability and Knowledge Reuse when is maximized (because a better surface finishingincreases production lead time and requires expert workers to be executed). Neutral points andtolerances (only for “optimize” functions) allow calculating DM scores for each value driver, which arethen aggregated (using a normalized weight based of the criticality of each driver) to obtain the totalDM for the IMC.Tab. 1: Extract from the CODA matrixComputer-Aided Design & Applications, 10(a), 2013, bbb-ccc 2013 CAD Solutions, LLC, http://www.cadanda.com

63PREFERENCES FOR VALUE VISUALISATION: RESULTS FROM THE EMPIRICAL STUDYThe empirical study took place between May 2009 and October 2012. In this timeframe the authors hadaccess to several aerospace companies, which contributed to the clarification of the problem domain,to the definition and validation of value assessment process, and to the development of thevisualization approach. The authors participated to regular physical workshops, virtual work meetingsand debriefing activities, where the findings were iteratively discussed and validated with the projectpartners. The major findings of the problem exploration phase, with regards to “what” to visualize and“how” to visualize value, are described in this section. These findings have been analyzed in the lightof the available literature, and used to develop a demonstrator of the value visualization approach.3.1What to Visualize?In spite of the centrality of the value concept in complex systems design, there is still relatively littleknowledge about what value is, what its characteristics are and how stakeholders determine it. Theneeds analysis moved from major value-oriented methodologies in engineering design, such asTradespace exploration [31], Value Engineering [15], Value Driven Design [12], to highlight gaps in theway value is currently visualized in Stage-Gate processes.When using value-oriented approaches in Systems Engineering, it is important for value functionsto be intuitive, meaningful, and allowing for direct comparisons among alternatives [13]. For thisreason monetary units are proposed as the most convenient, practical, and universally understoodmetric for value [12]. However, monetizing value attributes is cumbersome and meaningless inpreliminary design, and can potentially impede the timely communication of critical informationbetween all the relevant stakeholders [37]. A preference toward using simple scalars to rank designshas clearly emerged from the study. Scalars enable direct comparisons of heterogeneous drivers,putting the focus not only on physical and functional architectures, but also on relationship-basedaspects [37]. In addition, more than in producing an absolute value score, there is a preference inunderstanding how a concept is positioned against relevant benchmarks, to highlight if and how mucha solution is better or worse than reference options. Two main references have been identified at thispurpose: a product baseline (derived from historical data) and a target (which expresses a visionemerging from long-term forecasts).A major problem with value assessment in a preliminary phase, is that value models vary a lot interms of quality and reliability. This degree of uncertainty needs to be handled by assisting designersand decision makers in achieving a better understanding of what these uncertainties, ambiguities, andassumptions actually involve. In other words designers need to know which is the level of maturity[25] of the knowledge upon which the value models are built. The empirical study showed how criticalit is to have pointers that indicate the extent to which people should trust the material entering in thevalue assessment activity. Models able to communicate reliability and maturity of this informationhave been advocated as a major enhancement to support the communication of value information.3.2How to Visualize Value?Value is not the only criterion, but rather one of many criteria, upon which designers and processowners make decisions in a Stage-Gate process. A risk is that value contribution remains hidden byother factors, such as technical performances and cost. On the opposite side, value considerationsmight cause designers to lose focus on contractual requirements, causing non-conformities and delays.The empirical study has showed the need to merge heterogeneous information into a single interactive,visual environment, to facilitate decision-making activities, both in the Stage and at the Gate.Nowadays, product and process data are presented to the user with a collection of methods, suchas indented tree lists, reports and charts, which were found not to be suitable means for displayingvalue-related information Rather, the case study showed a preference in coupling such informationwith the assembly (or part) undergoing redesign. On the one hand, value-robust solutions can be moreeasily recognized if placed in the right context and related with the other design information. On theother hand, the implementation of 3D representations in a CAD/PLM tool allow for a streamlinednavigation through the assembly/product structure, facilitating information consumption andinterpretation.Computer-Aided Design & Applications, 10(a), 2013, bbb-ccc 2013 CAD Solutions, LLC, http://www.cadanda.com

7A preference has also been expressed in using multiple cues to encode value-related informationin the existing project documentation, to take advantage of associative processing [33]. Multiple cuesoffer more opportunities for the learner to discern the new information being presented [33], andprovide information quickly and automatically, decreasing time and effort needed to complete a task[5]. In this way designers can increase their perception of areas that are perceived low in terms ofvalue contribution and make corrective actions before the gate.The case study findings also show an inverted U-shaped relationship between the efficiency of thedecision making task and the amount of value information provided, a phenomenon already observed,in a more general context, by Zahakis and his team [41]. Hence, dealing with 5 or 6 main value criteriamay result to be as effective as using 200, but considerably less effective than using 25-30. In ordernot to overwhelm decisions makers with unnecessary details, the number of criteria to be presented ata time during the evaluation should kept to a minimum, while supporting information drilldown foraccess to more detailed data.4TRANSLATING VALUE MODEL RESULTS INTO COLOR-CODED FEATURES IN 3D CAD MODELSColors have emerged as one of the key cues for representing value-related information because of thebeneficial effects for decision-making observed and discussed since the 70s. Colors are among themost effective coding technique for aiding visual search [10][29]. The processing of color precedes theprocessing of other attributes [26] and is highly associative [28]. Colors support learning, as theyunderline figure ground relationships, interrelatedness and discrimination [11]. Also, subjects withcolor-coded reports have been observed to obtain a significantly higher average profit in less timewhen performing working tasks [6].The beneficial effects of color-coding have suggested the authors to map the value model resultsagainst a color scale, and to associate the resulting color to each relevant part/feature of the 3D CADassembly. The aim is to create a constant link between value information and product model, and, inthis way, facilitate trade-off analysis and the benchmarking of value dimensions different in nature.As stated in section 3.1, the empirical study has shown a preference towards benchmarkingalternative concepts against a baseline and target design. Hence the value model results are notdirectly translated into colors. Rather, the first step in the visualization consists in normalizing theoutcomes of the value model using the baseline and target references. One way is to map suchoutcomes against a numerical scale, which express the relative value contribution of a design concept.The numerical scale chosen is a 9-point scoring system that, given a design option #n, translates thevalue model result – the Design Merit (DMn) – into a value score (Sn) using Eqn. (1):Sn (St Sb )(DM n DMb )(DM t DMb ) Sb(1)where (DMb) represents the Design Merit computed for the baseline and (DMt) for the targetdesign. Similarly, (Sb) represents the score for baseline, which a-priori set equal to 3, while (St)represents the value score of the target, which is a-priori set equal to 8. It has to be noted that theformula is only applicable when (Eqn. 2):(7DMb 2DM t )(St Sb ) DM n DM t(2)On the one hand, in case (Eqn. 3):DM n DM t(3)the algorithm automatically assigns a score of 9 to the design alternative. Sn 9 denotes a designbetter in value compared to what was considered as the best desirable outcome for the forthcomingsolution. On the other hand, in (Eqn. 4):Computer-Aided Design & Applications, 10(a), 2013, bbb-ccc 2013 CAD Solutions, LLC, http://www.cadanda.com

8DM n (7DMb 2DM t )(St Sb )(4)the algorithm automatically assigns a score of 1. Sn 1 denotes a design scoring significantly belowthe baseline.To exemplify the use of these formulas, let’s consider a design concept that renders, as result of avalue modeling activity, a DMn 65%, while Baseline and Target render DMb 50% and DMt 70%.Running Eqn. (1) with these numbers, the team gets a score of Sn 6.75 for the concept under analysis.In another example, the team might obtain the same result (DMn 65%) from the value model, butdifferent merits for baseline and target (DMb 73%, DMt 85%), which render Sn 0.31. The empiricalstudy showed, however, that the design team is not interested in knowing how much a design is lower(or higher) than existing benchmarks, but rather just to know if the value contribution of a conceptfalls outside the boundaries. These upper and lower boundaries are calculated using Eqn. (2). It has tobe noted that, dealing with the lower boundary, the algorithm accommodates situations where thevalue of a design is slightly below what expressed by the baseline, to compensate for the lack of exactinformation when building and populating the value model. Hence, Eqn. (1) is applicable only if DMnis higher than 68.2% and lower than 85%. Given that DMn 65% 68.2%, Eqn. (4) assigns by default avalue score Sn 1 to this concept.The algorithm essentially renders four main areas:-Sn 1/2 indicates NO-GO designs, whose value contribution is below the baseline. Based on-Sn 3/4 indicates designs that meet the minimum threshold. This score might be-Sn 5/6/7 indicates designs in the GO area, although attention has to be paid on the-Sn 8/9 indicates designs with a value equal or higher than what was originally intended.the criticality of the value driver, this may cause the design to be definitively killed for notsatisfying such a minimum threshold. Otherwise, if the criticality is low, engineers mayaccept a lower value if this allows for more important dimensions to be improved.considered satisfactory if the criticality is low and major improvements have beenachieved in dimensions with higher priority. For more critical aspects, it may trigger thedecision to kill the design, especially when resources for rework activities are limited.reliability of the value assessment results. The design is moving in the right direction, butsome refinements may still be made to achieve the desired value outcome.Engineers can further analyze such over-the-target dimensions to trade-off excellentcapabilities with other drivers that are performing poorly, being free to decrease the valueof the first in order to increase the value of the latter.Value models are run for each part of an assembly. These parts are then color-coded according tothe scores obtained from the algorithm. The use of basic colors has not been preferred, becauseexperiments have shown that they do not segregate “exceptionally well” [36]. Rather, for specificapplications, chromatic gradation within hue or color category may be more appropriate [36]. Hence,the color scale selected features different color nuances ranging from red (lowest value contribution,Sn 1) to green (highest value contribution, Sn 9). Nuances of red indicate parts where the valuecontribution is equal or below the baseline design. Nuances of green indicate a value contribution inline with or above the target. Nuances of orange and yellow represent areas where the valuecontribution is above the baseline, but not yet satisfactory for the purpose of the project.5DEMONSTRATION WITH SIEMENS HD3D VISUAL REPORTING The value visualization approach has been prototyped and demonstrated in SIEMENS TeamCenter/NXusing HD3D Visual Reporting , a visually-rich environment for working with PLM data. NX HD3D VisualReporting provides an intuitive approach to report and conditionally format computational data in theComputer-Aided Design & Applications, 10(a), 2013, bbb-ccc 2013 CAD Solutions, LLC, http://www.cadanda.com

93D CAD model. Its characteristics have been found particularly appealing to couple the value modelresults to visual features, and to foster communication and management of the value information.The activation of NX HD3D Visual Reporting results in a conditional formatting (color-coding) ofthe assembly according to the particular data criteria under examination. The color-coding remains inforce for as long as the report is activated, allowing the user to work on objects of interest withoutlosing focus on the report results. Fig. 4 shows an example of color-coded IMC model for a relevantvalue driver (Availability). The results of the value studies conducted on each part of the assembly arefirst translated into colors, and then associated to the geometrical model. Aggregating the resultsobtained for each part, it is possible to obtain the overall value contribution of the IMC assembly(using an appropriate weighting algorithm).Fig. 4: Color-coded visualization of the IMC main parts (in SIEMENS NX HD3D Visual reporting).Value Drivers and their interrelationships are managed in SIEMENS Teamcenter as requirementsubtypes. Requirements are ordinary Teamcenter items individually organized in a hierarchicalstructure. Requirements include system defined attributes (e.g., identifier, time-stamps, userinformation, name, text), as well as a specific set of user defined attri

3Siemens PLM Software, henk.broeze@siemens.com 4Siemens PLM Software, gilles.dubourg@siemens.com 5Siemens PLM Software, clive.sandhurst@siemens.com ABSTRACT Recent literature in Systems Engineering has suggested the use of "value" to drive decision-making activities during preliminary design, in particular when choosing